Applying supervised intelligent scenarios to numerical investigate carbon dioxide capture using nanofluids

Abstract

Nanofluids have recently been engaged in absorption-based processes to capture carbon dioxide (CO2) molecules. Pressure, temperature, nanoparticles type and dosage in host liquid, and gas-liquid contact time determine the CO2 absorption capacity of nanofluids. An accurate approach is required for the feasibility study, preliminary design, and optimization of any potential nanofluid-based process for CO2 removal. This study first constructs different artificial intelligent models and then compares their prediction performances to introduce the most reliable one. All models are developed and validated using 460 experimental datasets. The adaptive neuro-fuzzy (ANFIS) equipped with the Sugeno-type inference system (cluster radius = 0.5) is the most accurate model for the given purpose. The engineered ANFIS model simulates the absorption capacity of pure water and water-based nanofluids with the lowest uncertainty (i.e., AARD = 2.09%, MSE = 1.9 × 10−4, R = 0.99691). ANFIS predictions showed that the Fe3O4 nanoparticles enhance the CO2 absorption capacity of water more than the Al2O3, SiO2, and carbon nanotubes.